Synthesis, evaluation, and computational chemistry of novel selenenyl sulfides as 3C protease inhibitors with strong cell-based antiviral activity

2.1 Chemistry of the compounds

Aiming at investigating the influence of aliphatic groups (R¹) and aromatic groups (R³) in the compounds, two different series, 7a-7u and 11a-11u were synthesized. The chemical routes are shown in Scheme 1 and Scheme 2, respectively. Although both compounds 7a-7u and 11a-11u were prepared via nucleophilic substitutions, corresponding reactions for these two series were different. Generally, 7a-7u were synthesized using a similar method to our previous work (shang et al., 2023), and a modification was that all intermediates 5 were synthesized from 4 in the presence of selenium powder, nano-Copper(I) iodide, and other related reagents. In comparison, compounds with close structure to 11a-11u have not been reported and the chemistry for the preparation of these compounds will be discussed. The starting material 8 (methyl 2-aminobenzoate or ethyl 2-aminobenzoate) was made into a diazonium salt 9 at iced temperature, after that potassium selenocyanate was added to the reactant in an acidic solution to give intermediate 10 (methyl 2-selenocyanatobenzoate or ethyl 2-selenocyanatobenzoate) in high yields, based on a known procedure (Diesendruck et al., 2009). Finally, 11a-11u were prepared by reacting 10 with different aromatic thiols 6 using aluminum oxide as a catalyst in dichloromethane at room temperature. The yields for 7a-7u were in the range of 7–85 % and those for 11a-11u were from 14 % to 67 %, respectively. Besides ¹H NMR and ¹³C NMR, all the selenenyl sulfides were also characterized by ⁷⁷Se NMR that is unique for the organoselenium compounds. The spectra of ¹H NMR, ¹³C NMR, ⁷⁷Se NMR, and HRMS for all the target compounds could be found in the supplementary data. Interestingly, for compounds from both 7 and 11 series, there existed a trend for the isomer compounds with only ortho or para position difference when R₂ group was a substituted phenyl ring, that the ortho substituted compound had a smaller ⁷⁷Se NMR chemical shift than the corresponding para substituted compound. Such phenomena could be observed from the following pairs of selenenyl sulfides: 7f (597.0) versus 7 g (566.7), 7i (597.8) versus 7 m (573.1), 7o (594.9) versus 7p (571.2), 11a (602.9) versus 11f (567.1), 11b (615.7) versus 11 g (594.1), 11c (597.0) versus 11 h (561.0), 11d (594.3) versus 11i (561.8), 11n (597.3) versus 11r (558.7), and 11o (594.2) versus 11 s (562.0), which was in agreement with the results in a previous study (Shang et al., 2023).

Close

Scheme 1

Reagents and conditions: (i) Reflux; (ii) Et3N, DCM, 0 °C to r.t.; (iii) Se, t-BuOK, nano-CuI, 1,10-phenanthroline, DMF, 110 °C; (iv) DCM, r.t.

Export to PPT

Close

Scheme 2

Reagents and conditions: (v) a. HCl, NaNO₂, 0 °C; b. CH₃COONa, 0 °C to r.t.; (vi) KSeCN, H₂O, r.t.; (vii) Al₂O₃, DCM, r.t.

Export to PPT

In order to confirm the chemical structures in the series of 11a-11u, single crystals of 11 s (CCDC number 2172212) suitable for X-ray diffraction were successfully obtained using a self-evaporation method in a mixed solution of hexane/dichloromethane(v/v = 100:1), after attempts on a few compounds. From Fig. 2, it can be seen that the molecule adopted an extended conformation in the crystal, from which a Se-S bond was successfully constructed between two phenyl rings and the ester part was linked with the left phenyl ring neighboring to the selenium atom (Se1 in the figure).

Close

Fig. 2

Crystal structure of compound 11 s.

Export to PPT

Simply from the view of chemistry, although limited compounds in the two series of selenenyl sulfides were synthesized in this study, more compounds with different structures could be considered for investigation. For example, the introduction of iodine atom to the structures might be an interesting topic, as previously shown as halide system (Lun et al., 2023).

2.2 Biological activity

The target selenenyl sulfides were all subjected to enzyme inhibitions against EV71-3C^pro and CVB3-3C^pro, and in vitro cell-based evaluations against these two viruses, the results of which have been summarized in Table 1. Generally, most compounds exhibited potent inhibitions against 3C^pro and ideal in vitro antiviral activities. In order to filter those with very weak inhibitions, the cutoff of IC₅₀ value was set to 10 μM for 3C^pro and that was 30 μM for the cell-based activity. For comparison, rupintrivir was chosen as a control in this study.

Amongst the compounds, 7b, 7c, 7d, 7f, 7 g, 7i, 7 l, 7o, 7p, 7q, 7r, 11a, 11b, 11e, 11f, 11 g, 11j, 11 m, 11o, 11q, 11 t, and 11u had IC₅₀ values < 10 μM for both proteases; 7a, 7 s, 7 t, and 11 k had IC₅₀ values < 10 μM for EV71-3C^pro; and 7j, 7k, 7m, 11c, 11d, 11h, 11i, 11n, 11p, 11r, and 11 s had IC₅₀ values < 10 μM for CVB3-3C^pro, respectively. While for 7e, 7h, 7n, 7u and 11 l, both the IC₅₀ values against EV71-3C^pro and CVB3-3C^pro were > 10 μM. For the EV71-3C^pro inhibitions, 7b, 7l, 7 t, 11g, 11j, and 11u and 11f had IC₅₀ values < 2.5 μM, among which 11f had the best IC₅₀ value of 0.28 ± 0.05 μM, much stronger than that of rupintrivir (1.03 ± 0.53 μM). For the CVB3-3C^pro inhibitions, 11f, 11g, 11h, and 11u had IC₅₀ values < 2.50 μM, comparable to that of rupintrivir (2.01 ± 0.22 μM). Fig. 3a and 3b. display the inhibition curves of 11f against EV71-3C^pro and CVB3-3C^pro. Regarding a simple structure–activity relationship, when R² was a 4-nitrophenyl group for compound 7, two of the compounds displayed the best inhibitions against EV71-3C^pro (7b and 7 l). However, this group did not give strong enzyme inhibitions for 7u, as well as the corresponding compounds for 11 (11e and 11p). For compound 11, when R² was a 2-methylphenyl group, 2-fluorophenyl group or 3-trifluoromethylphenyl group, potent enzyme inhibitions were observed (11f, 11 g and 11j). However, 3-trifluoromethylphenyl group in 7 h showed very poor enzyme inhibition. When R² was a naphthalene-2-yl, 11u gave strong inhibitions for both 3C^pro,

Close

Fig. 3

Inhibition curves of 11f against EV71-3C^pro (a) and CVB3-3C^pro (b).

Export to PPT

nevertheless, 11 l exhibited no activity even at 10 μM.

It is a concern whether the target selenenyl sulfides inhibit the 3C protease through a covalent bonding manner. From the previous study, the aromatic disulfides did not form a covalent bond with the cysteine residue of SARC-CoV-3CL^pro in the active site, which was evidenced by that fact that mass spectrometry of the protease had not been changed and the nonsymmetrical disulfides were reversible inhibitors (Wang et al., 2017). In another study, the ebselen-cysteine adduct was not found with Cys145 of SARS-CoV-2-3CL^pro, using LC-MS/MS analysis of the tryptic digest of the enzyme, although such an adduct was observed in the ebselen-treated human glutathione S-transferase-pi (Amporndanai et al., 2021). In this study, inhibition type of the 11f was further determined by means of enzymatic kinetic behaviors, from which a competitive inhibition behavior was observed and it implied the inhibition was a reversible action (Fig S1). Due to all these reasons, it is quite likely that the selenenyl sulfides do not form covalent bond with the cysteine in the active site of the 3C proteases.

Based on the cell-based evaluation, none of 7 h, 7i, 11j, 11 l, and 11n had good activities against both EV71 and CVB3, 7a, 7 g, 11o, 11p, 11 k, 11 s and 11 t had one IC₅₀ value > 30 μM against either EV71 or CVB3, and all the other selenenyl sulfides showed IC₅₀ values of < 30 μM against the two viruses. For the cell-based activities against EV71, 11f and 11u displayed IC₅₀ values of < 2.50 μM, which were the best ones among all the compounds but still weaker than that of rupintrivir (0.74 ± 0.22 μM). For the antiviral activities against CVB3, the IC₅₀ values of 11e and 11f were 1.15 ± 0.04 μM and 0.72 ± 0.15 μM, respectively, which were several folds stronger the value of rupintrivir (6.52 ± 0.09 μM). Fig. 4a shows the dose–response curve of 11f against CVB3. Although the cell-based assay is a simple in vitro model, it is obvious that the properties of absorption, distribution, metabolism and excretion (ADME) should be taken for consideration for the cellular activity, since the cellular environment is much more complex than the pure enzyme inhibition. Therefore, it was not strange that there did not exist an absolute correction between the enzyme inhibition and the antiviral activity for the tested selenenyl sulfides. On the other hand, there exists such a possibility, that the antiviral activity of these selenenyl sulfides might involve targeting other viral proteins or host proteins in addition to 3C^pro_. This is due to a fact that ebselen and some other compounds did not inhibit the viral replication of EV-A71, although those compounds inhibited cysteine proteases significantly (Ma et al., 2020).

Close

Fig. 4

(a) Inhibition curve of 11f against CVB3. (b) Anti-CVB3 activity of 11f from viral plaque assay (the symbol *** means the significant difference p-value is no more than 0.001).

Export to PPT

The antiviral activity was further supported by viral plaque assay, compound 11f showed very strong reduction of CVB3 particle formation activity at 5 µM, with 80 % inhibition as compared to DMSO controls (Fig. 4b). A subsequent cell cytotoxicity assay for 7q, 7r, 7 s, 11e, 11f, 11 g, 11u and rupintrivir against normal Hela cells revealed that selectivity indexes for these compounds were 6.6, 9.4, 8.2, 23.6, 51.5, 6.8, 7.0, and 7.6 for CVB3, respectively (Table 2). Taken together, 11f was identified as a potent antiviral agent targeting 3C/3CL^pro with advantageous bio-safety.

For a general comparison of the antiviral efficacies of between the two series 7a-7u and 11a-11u, compounds belonging to the second panel exhibited stronger cell-based antiviral activity. For the anti-CVB3 activity, the IC₅₀ values of 11e and 11f were around 1 μM, however, none of the compounds from the first series was at the same level. This implied that more attention should be paid to compounds with aliphatic groups (R¹) for the next round of molecular design to discover more potent compounds.

Regarding to whether the ester or the amide part of the selenenyl sulfides are stable in the cell-based assay, there is an assumption that the active component in the cellular environment is possible to be the hydrolyzed product. Nevertheless, this might not be true based on the following fact: 7b, 7 l, 7u, 11e and 11p had same R² groups, however, the antiviral activities of them differed significantly. Otherwise, these compounds tended to demonstrate similar antiviral activity. Further experiments will be conducted to study the components after the cell-based assay for a detailed investigation.

In a previous result, we screened a big library of disulfide compounds against CVB3-3C^pro and evaluated their cell-based anti-CVB3 activities. Although some compounds showed submicromolar IC₅₀ values against CVB3-3C^pro, the antiviral activities were very poor even at 10 μg/mL concentration (corresponding to approximately 30–40 μM based on different molecular weight of each compound), much weaker than that of rupintrivir at the same condition. That work was not published in a scientific paper, but shown in a patent application (Zhang et al., 2016). The present result suggested that although selenenyl sulfides are bioisosteres of the disulfide compounds, the former have more preferable biological behavior in the cell-based antiviral assay and have much greater potential for further development of novel antiviral agents.

Hundreds of papers have been published on the discovery of inhibitors of SARS-CoV-2 since the outbreak of COVID-19, however, there were very few reports on the identification of anti-CVB3 or anti-EV71 inhibitors. The preliminary result suggested that selenenyl sulfides in the present study were effective against CVB3 infection.

2.3 Computational chemistry

2.3.1

2.3.1 Proposed binding mode of active compound

Molecular docking has become a useful tool to predict the possible binding mode of inhibitor with biological target, when the co-crystal structure is unavailable (Bondock et al., 2023; Meng et al., 2022; Zhao et al., 2023;). To explore the plausible mode of action for this family of antiviral compounds, 11f was docked the active site of EV71-3C^pro (pdb code 3sjo) (Lu et al., 2011). This was because: (i) 11f was more potent than rupintrivir agaisnt EV71-3C^pro; (ii) Rupintrivir was in the binding pocket of this crystal structure; (iii) EV71-3C^pro and CVB3-3C^pro share fairly strong similarity. It should be noted that although this pdb file contained eight chains, these chains were identity in the sequence and the 3D structures superimposed one another perfectly. Therefore we choose chain A for the molecular simulation. Fig. 5a shows the interaction of 11f and the neighboring residues of EV71-3C^pro from the molecular simulation in comparison with rupintrivir in the crystal structure. As can be seen, 11f was much smaller than rupintrivir in size, and 11f overlaid with the left part of rupintrivin. The selenenyl sulfide inhibitor was involved in multiple hydrophobic contacts with His24, Phe25, His40, Lys143, Ala144, Gly145, His161, Ile162, Gly163, and Gly164, while Gln146 and Cys147 were found to form two hydrogen bonds with 11f. Fig. 5b depicts the three-dimensional molecular surface of the binding cavity generated by PyMOL and 11f adopted a relaxed conformation to inhabit in the cavity. Although fewer interactions were observed for 11f than for rupintrivir with the enzyme, much stronger inhibition of 11f against EV71 was determined experimentally. We will try the co-crystallization of this compound with EV71-3C^pro or CVB3-3C^pro to explain the exact mode of action in the future.

Close

Fig. 5

Predicted binding mode of 11f with EV71-3C^pro. (a) Comparison of 11f (lightblue) with rupintrivir (cyan), in which the full protein was shown in a cartoon model. (b) The molecular surface of the binding pocket (11f shown in stick model and colored by element).

Export to PPT

2.3.2

2.3.2 Structure-activity relationship

Comparative field analysis (CoMFA) is powerful to analyze the quantitative structure–activity relationships (QSAR) of bioactive compounds by steric and electrostatic contributions using three-dimensional models (Cramer et al., 1988). The conformation of 11f from molecular docking was used to build all the molecular structures. For this study, the cell-based IC₅₀ values against CVB3 were selected as the biological data to generate the CoMFA model because: (i) There was a bigger training set for building the model; (ii) CoMFA is not limited only to analyze the enzyme inhibitory data. Besides 7 h, 7i, 11j, 11 l, 11n and 11 s, the IC₅₀ values of which were > 30 μM, compounds 7u, 11 t and 11u were also excluded from the training set because they were statistical outliers. The final training set gave a leave-one-out q² of 0.519 when the number of optimum components was 6. The non-crossvalidated r² was 0.931 with a standard error of estimate of 0.107 and F values of 18.118. This statistical result indicated that the CoMFA model was successful, even if the compounds belongs to two slightly different series. The steric contribution was 67.3 % and the electrostatic contribution was 32.7 %. The predicted biological data and experimental biological data from CoMFA were listed in Table 3. The most active compound 11f was used to explain the contour maps, as shown in Fig. 6. For the steric map (Fig. 6a), a bulky group in the green space is favorable for anti-CVB3 activity, and such a group in the yellow region is likely to decrease the antiviral activity. For the electrostatic map (Fig. 6b), an increase of the positive charge will give enhanced activity in the blue region, whereas in the red space, negatively charged group is likely to improve the cell-based inhibition. Inspired by the above computational chemistry, we will make further structural optimization to seek for compounds with improved activity. For example, it is practical in organic synthesis to introduce a halogen atom at 4-position of the phenyl ring that attached to the sulfur atom of 11f, and to change the ester moiety to ethyl or propyl group. The next round of molecular design is currently underway.

Table 3 Predicted biological data versus experimental biological data from CoMFA.

Compound	ED	PD	ER	Compound	ED	PD	ER
7a	4.77	4.58	0.19	7 t	5.02	4.97	0.05
7b	4.97	5.06	−0.09	11a	4.91	5.02	−0.11
7c	4.83	4.86	−0.03	11b	4.95	4.81	0.14
7d	5.03	5.00	0.03	11c	4.61	4.80	−0.19
7e	4.58	4.70	−0.12	11d	4.88	4.73	0.15
7f	4.75	4.65	0.10	11e	5.94	5.99	−0.05
7 g	4.60	4.58	0.02	11f	6.14	6.15	−0.01
7j	4.85	4.82	0.03	11 g	5.28	5.21	0.07
7 k	4.63	4.65	−0.02	11 h	4.78	4.83	−0.05
7 l	5.18	5.22	−0.04	11i	4.88	4.89	−0.01
7 m	4.72	4.75	−0.03	11 k	5.14	5.38	−0.24
7n	4.80	4.82	−0.02	11 m	4.98	4.90	0.08
7o	4.83	4.92	−0.09	11o	4.58	4.65	−0.07
7p	4.84	4.85	−0.01	11p	4.75	4.78	−0.03
7q	5.20	5.18	0.02	11q	4.84	4.93	−0.09
7r	5.27	5.19	0.08	11r	4.66	4.54	0.12
7 s	5.20	5.14	0.06	11 t	4.74	4.54	0.20

*ED = experimental pIC₅₀ data, PD = predicted pIC₅₀ data, ER = error.

Close

Fig. 6

Three-dimensional contour maps of the anti-CVB3 activities from the CoMFA model for (a) steric field and (b) electrostatic field contributions.

Export to PPT

4.1 Chemistry

The reagents and solvents for the chemical preparation were all of analytical grade quality, purchasing from the online mall for laboratory use in Nankai University (mall.nankai.edu.cn/labmai/cloudbean). All the liquid reagents and organic solvents were dried in advance and distilled using standard methods before use. All the reactions were conducted using oven dried glassware. Melting points were determined using an RT-2 melting apparatus (Shanghai PuZhe photoelectric Co., China) and were uncorrected. ¹H NMR, ¹³C NMR, and ⁷⁷Se NMR spectra were recorded using a Bruker Avance 400 MHz spectrometer (Bruker Corporation, Switzerland). The chemical shift data (δ) for the nuclear magnetic resonance spectra were expressed as parts per million (ppm), using deuterated chloroform (CDCl₃) or dimethyl sulfoxide (DMSO‑d₆) as solvent and tetramethylsilane (TMS) as an internal standard reference. High-resolution mass spectra were recorded on an FT-ICR mass spectrometer (Ionspec, 7.0 T) or a Varian QFT-ESI instrument (Varian Medical Systems, Crawley, UK). Single crystal X-ray diffraction was carried out on a Bruker Smart 1000 CCD diffractometer (Bruker Corporation, Switzerland). General silica gel (200–300 mesh) was used for the column chromatography purification.

4.2 General procedure for synthesis of intermediates and target compounds 7a-7u

4.3 General procedure for synthesis of intermediates and target compounds 11a-11u

4.4 Biological reagents and conditions

Sodium chloride, potassium chloride, disodium hydrogen phosphate dodecahydrate, potassium dihydrogen phosphate and dimethyl sulfoxide were purchased from National Pharmaceutical Group Chemical Reagent Co., Ltd. Tris-(hydroxymethyl)-aminomethane was purchased from Novon Scientific. Imidazole was purchased from MP Biomedicals. Dithiothreitol was purchased from Beijing Xinjingke Biotechnology Co.,Ltd. Bovine serum albumin and glycerol were purchased from Sigma. Rupintrivir was purchased from Toronto Research Chemicals. Cell counting Kit-8 was purchased from Dojindo Chemicals. Substrate for the 3C^pro/3CL^pro_, H-Asp-{Glu(EDANS)}-Met-Ser-Ala-Ile-Phe-Gln-Gly-Pro-Ile-Ser-{Lys(DABCYL)}-Asp-OH was purchased from GL Biochem in Shanghai. 96-well microtiter plates were purchased from Corning Co. Ltd. Enzyme inhibition and cell based activity was recorded on a Spectra max 190 microplate reader (Molecular Devices Co. Ltd.). All the experiments were conducted in a standard biosafety shelter laboratory 2 (BSL-2).

4.5 Enzyme inhibition assay

EV71-3C^pro and CVB3-3C^pro were expressed and purified similar to the previous method (Lu et al., 2011; Wan et al., 2023). The DNA encoding EV71-3C or CVB3-3C protease was amplified and inserted into pET28a (Novagen). The resulting plasmid encodes an N-terminal His-tagged EV71-3C or CVB3-3C protease. For the expression and purification, the plasmid was transformed into Rosetta™ (DE3) Competent Cells (Novagen). Bacteria culture was grown in LB culture at 37 °C until the density OD₆₀₀ reached 1.0, after that IPTG was added to the culture (final concentration of 0.5 mM) to induce expression, which was shaken for another 20 h at 18 °C. Bacterial cells were collected by centrifugation at 4000 g, and resuspended in the lysis buffer that contained 50 mM Tris-HCl (pH 8.0), 150 mM NaCl and 20 mM imidazole. After that the bacterial cells were disrupted by ultra-sonication and cell debris was removed by centrifugation at 20,000 g for 1 h before the supernatant was loaded to Ni-nitrilotriacetic acid resin (GE Healthcare, USA) pre-equilibrated with the lysis buffer. Nonspecific contaminants were further removed by washing the resin with 10x bed volume of the wash buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 20 mM imidazole. For the next step the recombinant 3C proteins were eluted with the elution buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 300 mM imidazole. 6xHis-tag was cleaved by Thrombin (Sigma). In the end 3C protease was purified by gel-filtration (Superdex 200 column, GE Healthcare, USA) pre-equilibrated with the gel-filtration buffer that contained 25 mM HEPES (pH 8.0), 150 mM NaCl. Fractions containing 3C protease were pooled and concentrated to 5 mg/mL.

CVB3-3C^pro inhibition assay was taken as an example here and the assay for EV71-3C^pro was highly similar. The tested compounds were dissolved in DMSO stock solutions of 50 mM concentration. The total volume of the reaction buffer was 50 μL, which included 2.3 μM of CVB3-3C^pro, 25 μM of substrate and PBS-BSA buffer (PBS buffer × 2 with 2 mg/mL BSA and 5.85 mg/mL NaCl), and 0–200 μM of the tested compounds of double concentration gradient. Rupintrivir was selected as a control for comparison. The reaction was incubated at 37 °C for 60 min, followed by the fluorescence absorbance recorded by the microplate reader. The excitation/emission light was 340/490 nm. The IC₅₀ values were calculated by Eq. (1). v = v₀ / (1 + [I] /IC₅₀)(1)

where v is the inhibited rate, v₀ is the uninhibited rate, and [I] indicates the concentration of the inhibitor. All the assays were finished in triplicate.

4.6 Cell-based antiviral evaluations

The recombinant CVB3 and EV71 viruses were constructed using reverse genetic approaches from 12 plasmid system and restored in the Key Laboratory of Pathogenic Microorganisms and Immunology, Institute of Microbiology, Beijing. Good conditioned Hela cells were cultured in 12-well plates and incubated at 37 °C and 5 % CO₂ environmental condition for 24 h, and then the culture medium was changed to DMEM medium for 1 h. After that the inhibitors were added to the medium in different concentrations and incubated for 40 min. Subsequently, the Hela cells were infected by CVB3 or EV71 with a multiplicity of infection (MOI) of 0.1. The positive control only contained the Hela cells and did not contain the inhibitor and the virus, and the negative control contained the Hela cells and the virus. A further 16 h incubation was performed until the negative control exhibited obvious cytopathic effect (CPE) and then the cell viability was measured using the MTS reagent. The IC₅₀ values were calculated by equation (2). IC₅₀ = [C_L(I_H −50) + C_H(50 - I_L)] / (I_H - I_L)(2)

where C_L means low concentration and I_L means inhibitory data at this concentration, while C_H means high concentration and I_H means inhibitory data at this condition. All the assays were evaluated in triplicate.

Viral titer was determined by the viral plaque assay. In detail, when Hela cells reached 95 % density, the cells were infected by different concentrations of CVB3 for 1 h. After washed by PBS, DMEM containing 1.5 % low melting point agarose and 2 mg/L pancreatic enzyme was added. After 3 d incubation, the viral plaque was counted.

The cytotoxicity assay was similar to the above cell-based antiviral experiment, and the only difference was that the Hela cells were not infected by the virus.

4.7 Molecular docking

The molecular structure of the 11f was constructed within Sybyl 7.3 (Tripos Inc., St Louis, MO), based on the crystal structure of 11 s. The structure model was assigned Gasteiger-Hückel charges and minimized by the Tripos force field. FlexX module was used to carry out the molecular docking task. The crystal structure of EV71-3C^pro in complex with rupintrivir (pdb code 3sjo) was retrieved from the pdb databank. All water molecules were removed, and hydrogen atoms were added in the standard geometry. Any amino acid residue within 5.0 Å of the location of rupintrivir was considered to be in the binding pocket. Cscore calculation was enabled and set to serial mode. The scoring procedures were performed using the default parameters in the program.

4.8 Comparative field analysis

All the 3D structures of the compounds were built based on the structure of 11f from molecular docking result and treated using same manner within Sybyl 7.3. All the parameters were used the default value within CoMFA module and the column filtering was set to 2.0 kcal/mol. The “leave-one-out” (LOO) cross validation method was tried to determine the optimum number of partial least squares (PLS) components. Finally, the non-cross validated method was used to derive the model to explain the quantitative structure-activity relationship.